CN115880501A - High-voltage wire infrared image processing method and system based on infrared camera - Google Patents
High-voltage wire infrared image processing method and system based on infrared camera Download PDFInfo
- Publication number
- CN115880501A CN115880501A CN202211503387.3A CN202211503387A CN115880501A CN 115880501 A CN115880501 A CN 115880501A CN 202211503387 A CN202211503387 A CN 202211503387A CN 115880501 A CN115880501 A CN 115880501A
- Authority
- CN
- China
- Prior art keywords
- image
- temperature
- pixel
- picture
- point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a high-voltage wire infrared image processing method and system based on an infrared camera, wherein after a single picture is marked according to points, lines and rectangles, information such as the highest temperature, the lowest temperature and the average temperature of a corresponding marked area is obtained after the mark is clicked on the picture; the line mark is used for displaying a temperature point trend graph on the line segment in the temperature trend display module; the pictures are automatically processed after the early warning temperature and the threshold parameter are set through the picture batch processing function, the pictures with abnormal and normal temperatures and not meeting the processing requirements are separately stored, and the current processing progress is prompted through a progress bar; the function of providing an interactive interface for the infrared image shot by the unmanned aerial vehicle is realized, the operation is simple, the function is visual, and the operation is stable. After the source folder is selected, the method can quickly detect the areas of the conducting wires and the drainage wires in the picture. The invention is friendly to interaction, can quickly respond to various instructions and visually display the final processing result.
Description
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to a high-voltage wire infrared image processing method and system based on an infrared camera.
Background
With the vigorous development of the unmanned aerial vehicle industry in China, the unmanned aerial vehicle has the advantages of being small and exquisite, flexible, high in flying height and the like, and a plurality of high-altitude shooting operations are carried out by the unmanned aerial vehicle. Nowadays, utilize unmanned aerial vehicle to carry the thermal imaging system, whether the temperature that detects high-tension line is normal to guarantee that the safety of high-tension line becomes an important technological means. However, for the collected picture, how to visually display and process the analyzed temperature information needs to be used, and a good interactive interface needs to meet the requirements of simple operation, visual function, stable operation and the like.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the high-voltage wire infrared image processing method and system based on the infrared camera are used for providing an interactive interface for an infrared image shot by an unmanned aerial vehicle.
The technical scheme adopted by the invention for solving the technical problems is as follows: a high-voltage wire infrared image processing method based on an infrared camera comprises the following steps:
s1: collecting an infrared image by an unmanned aerial vehicle thermal imager;
s2: preprocessing the infrared image to determine the visual characteristics of a linear target in the infrared image, and calibrating the object topology of an image pixel space; the method comprises the following specific steps:
s21: carrying out graying operation on the infrared image to obtain a corresponding grayscale image; then preprocessing the gray level image including denoising and smoothing; then judging and removing non-target objects through a connected domain of the gray level image, and extracting ROI areas of the leads and the drainage wires;
s22: performing morphological operation on the image according to the ROI of the lead and the drainage wire to connect the insulator connected regions into a whole; then searching the largest connected domain area in the image, namely the insulator part with the largest visual area in the image;
s23: in the gray scale image, counting the number of pixel points with higher gray scale value on the lines by taking the boundary of the insulator on the side of the wire as a starting point and the intersection point of the wire and the image boundary as an end point through the positions of the insulators and the number of the wires, wherein the line with the largest pixel points is the detected wire;
s24: extracting the position of the wire, sequentially calculating the gray values of pixel points on the wire in an original gray image to perform secondary positioning, eliminating error pixel points and obtaining accurate single-pixel wire position information;
s3: and performing targeted processing according to the actual characteristics of the linear objects in the image to finish semantic segmentation and target area refinement extraction of the linear objects.
According to the scheme, in the step S21, the specific steps are as follows:
s211: carrying out RGB channel separation on the infrared image to obtain a gray image;
s212: performing binary processing on the gray level image by adopting a self-adaptive gray level threshold algorithm for denoising the image;
s213: and extracting an ROI (region of interest) region of the denoised gray image by adopting a K-means algorithm, and being used for delimiting and segmenting the main body target contour of the linear target.
Further, in step S213, the specific steps are:
s2131: setting the number k of clusters of the gray image needing iteration by using the gray distribution frequency characteristics of the histogram and the number of peak valleys and slow areas;
s2132: setting the central point of the initial grouping by using a quantile point strategy, gradually realizing the mean value iteration in each unit, and finishing the segmentation;
arranging all the gray values which are not subjected to grouping processing in the processed gray image into a vector set X according to a monotone rising principle;
the set X is equally divided into k parts according to the set k value, thenThe position between each aliquot is the position P of the quantile i ：
S2133: calculating the optimal classification quantity by using an optimization criterion, verifying a preset k value until an ideal segmentation effect is met, and then deriving a segmented infrared image;
in a pixel set, the internal difference of each classified group is minimized, namely the numerical value of each element is as close to the average value of the group as possible, and the difference between the groups is maximized, namely the difference between the average values of different groups is larger, so that the segmentation effect is better;
the internal difference value of the same group is expressed as a standard deviation between the pixel values of all elements in one group and the pixel value corresponding to the central point, i.e. the average value within the group;
let n be the number of all pixels in the set, C i For the ith group, x is the pixel value of each pixel in the ith group, and Z is the average of all the pixel values in the ith group, then the internal difference value S of the whole pixel set in Comprises the following steps:
further, in step S22, the specific steps are:
s221: detecting straight lines in the image by adopting a Canny operator, and converting the detection of a set of points with more significant local changes in the gray level image into edge detection;
s222: and (3) performing connected domain processing by adopting a Seed-Filling Seed Filling method to obtain a maximum connected domain, wherein the connected domain is the position of the insulator.
Further, in step S221, the specific steps include:
s2211: let (x, y) be the coordinates of a certain pixel in the image, x 2 +y 2 Is a half of Gaussian blurAnd the diameter sigma is the standard deviation of normal distribution, a two-dimensional Gaussian function is adopted to carry out smoothing operation on the image to obtain a data array I (x, y) for inhibiting the adverse effect of image noise factors on the edge detection performance, and the two-dimensional Gaussian function is as follows:
in a two-dimensional space, the contour lines of the curved surface generated by the formula are concentric circles which are normally distributed from the center; a convolution matrix formed by pixels with distribution not equal to zero is transformed with the original image, and the value of each pixel is the weighted average of the values of the surrounding adjacent pixels; the value of the original pixel has a maximum Gaussian distribution value and a maximum weight; the farther the adjacent pixel is from the original pixel, the smaller the weight thereof;
s2212: calculating the gradient amplitude and the direction of the smoothed data array I (x, y) by selecting a method of solving finite difference mean values of all pixel points in a 2 x 2 neighborhood to obtain a corresponding gradient amplitude image and a corresponding gradient direction image;
the partial derivatives in the x-and y-directions are:
G x ＝(I(i,j+1)-I(i,j)+I(i+1,j+1)-I(i+1,j))/2
G y ＝(I(i,j)-I(i+1,j)+I(i,j+1)-I(i+1,j+1))/2；
the gradient magnitude and gradient direction are respectively:
s2213: carrying out non-maximum value inhibition, detection and edge connection by adopting a Canny algorithm to obtain an edge point array of the image, and carrying out further edge thresholding operation to remove false edges;
implementing a direction angle normalization operation in a gradient direction of the image to normalize to four angles;
the amplitude of all non-roof ridge peak values existing in the gradient direction is restrained, and the effect of thinning the gradient amplitude roof ridge in the amplitude array A [ i, j ] is achieved;
selecting a template with the size of a 3 multiplied by 3 pixel window and containing 8 direction neighborhoods to act on all points of the amplitude array A [ i, j ], comparing a central pixel with two pixels along a gradient line at each point, and setting the A [ i, j ] to zero if the amplitude A [ i, j ] at the central pixel point is not larger than the amplitudes of two adjacent points along the gradient line in the comparison process.
According to the scheme, in the step S23, the specific steps are as follows:
s231: determining the position information of the insulator according to the maximum connected domain;
s232: estimating the position of a boundary point of the wire at the edge of the image according to the inclination of the insulator;
s233: and determining the final position of the wire by counting high-gray pixel points between the end point of the wire at the insulator and the boundary end point.
According to the scheme, in the step S3, the specific steps are as follows:
s31: storing a plurality of images in the same folder into different result folders according to different temperature results of the same part;
s32: selecting different point marks, line marks and rectangular marks for a single image to mark in the image; the drainage lines, wires and corresponding maximum, minimum and average temperatures in a single infrared image are marked.
Further, in step S32, the specific steps include:
s321: the method comprises the steps of marking points, displaying temperature values of the points after a user clicks the point marks in an image, and displaying information including the highest temperature, the lowest temperature and the average temperature of the point marks in a temperature information display area after clicking the point marks for multiple times;
s322: the method comprises the steps that a line alignment mark is marked, a user clicks a starting point and an ending point on an image to generate and display a line segment, a temperature trend graph of the points on the line segment is displayed in a temperature trend display area, information including the highest temperature, the lowest temperature and the average temperature of the line segment is displayed in a temperature information display area, and specific positions of the highest temperature and the lowest temperature are marked in the image;
s323: for the rectangular mark, a user clicks a starting point and an ending point on the image to generate and display a rectangle, information including the maximum temperature, the minimum temperature and the average temperature of the rectangle is displayed in the temperature information display area, and meanwhile specific positions of the maximum temperature and the minimum temperature are marked in the image.
A high-voltage wire infrared image processing system based on an infrared camera comprises a picture folder selection module, a picture storage module, a folder directory tree-shaped display module, a picture processing display module, a mark type selection module, a temperature trend display module, a temperature information display module, a picture information display module and a picture batch processing parameter setting module;
the picture folder selection module is used for providing pictures for subsequent operations;
the picture saving module is used for saving the current picture according to the selected path;
the folder directory tree-shaped display module is used for selecting pictures in the picture folder according to the path;
the picture processing and displaying module is used for marking drainage wires and conducting wires and corresponding highest temperature, lowest temperature and average temperature information in the selected picture;
the mark type module is used for matching with the picture processing and displaying module and displaying corresponding maximum temperature, minimum temperature and average temperature information on the picture by clicking the mark;
the temperature trend display module is used for displaying a line segment on the picture according to the line marking instruction, the positions of the starting point and the ending point and displaying the temperature distribution trend of all points on the line segment;
the temperature information display module is used for displaying the highest temperature, the lowest temperature and the average temperature information of all the points on the picture according to the point marking instruction and the positions of the points; the temperature display device is also used for displaying the information of the highest temperature, the lowest temperature and the average temperature of each line according to the line marking instruction and the positions of the multiple lines, and displaying the temperature trends of different lines; the temperature display module is also used for displaying the highest temperature, the lowest temperature and the average temperature information of each rectangular area according to the rectangular marking instruction and the position of the rectangular area, and marking the position of the highest temperature and the lowest temperature in the corresponding rectangular area;
the picture information display module is used for displaying the related information of the current picture;
the picture batch processing parameter setting module is used for respectively storing pictures with the highest temperature meeting the sum of the early warning temperature value and the threshold value.
A computer storage medium having stored therein a computer program executable by a computer processor, the computer program executing a high voltage wire infrared image processing method based on an infrared camera.
The beneficial effects of the invention are as follows:
1. according to the high-voltage wire infrared image processing method and system based on the infrared camera, after a single picture is marked according to points, lines and rectangles, the highest temperature, the lowest temperature, the average temperature and other information of a corresponding marked area are obtained after the mark is clicked on the picture; the line mark is used for displaying a temperature point trend graph on the line segment in the temperature trend display module; the pictures are automatically processed after the early warning temperature and the threshold parameter are set through the picture batch processing function, the pictures with abnormal and normal temperatures and not meeting the processing requirement are separately stored, and the current processing progress is prompted through a progress bar; the function of providing an interactive interface for the infrared image shot by the unmanned aerial vehicle is realized, the operation is simple, the function is visual, and the operation is stable.
2. After the source folder is selected, the method can quickly detect the areas of the conducting wires and the drainage wires in the picture.
3. The invention is friendly to interaction, can quickly respond to various instructions and visually display the final processing result.
Drawings
FIG. 1 is an algorithmic flow diagram of an embodiment of the present invention.
FIG. 2 is a flow chart of a pre-process of an embodiment of the present invention.
Fig. 3 is a segmentation flow diagram of an embodiment of the present invention.
Fig. 4 is a diagram illustrating the ROI extraction effect according to the embodiment of the present invention.
FIG. 5 is a diagram illustrating the effect of edge detection according to an embodiment of the present invention.
Fig. 6 is a schematic diagram of the 4 and 8 neighborhoods of the embodiment of the present invention.
FIG. 7 is a flow chart of a seed filling algorithm of an embodiment of the present invention.
Fig. 8 is a diagram of the maximum connected region of the insulator according to the embodiment of the present invention.
Fig. 9 is a flow chart of secondary positioning according to an embodiment of the present invention.
Figure 10 is a diagram of the location of detected insulators and found conductors in accordance with an embodiment of the present invention.
Fig. 11 is an overall block diagram of infrared image processing software according to the embodiment of the present invention.
Fig. 12 is a diagram illustrating an effect of selecting a picture folder by clicking a file button in a menu according to an embodiment of the present invention.
FIG. 13 is a diagram illustrating an effect of clicking on a file directory tree to display a picture according to an embodiment of the present invention.
Fig. 14 is a graph showing the effect of the running algorithm of the embodiment of the present invention finding the drainage wire and the conducting wire, displaying the temperature information, and identifying the highest temperature and the lowest temperature on the picture.
Fig. 15 is a diagram showing the effect of selecting a point marker and displaying temperature information according to the embodiment of the present invention.
FIG. 16 is a graph of the effect of selecting line markers and displaying temperature information according to an embodiment of the present invention.
Fig. 17 is a diagram illustrating an effect of selecting a rectangular mark and displaying temperature information according to an embodiment of the present invention.
Fig. 18 is a parameter setting diagram of batch processing pictures according to the embodiment of the present invention.
FIG. 19 is a diagram illustrating the effect of a batch run according to an embodiment of the present invention.
Fig. 20 is a diagram showing the effect of the end of the classification of the batch processing according to the embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
For infrared images taken by a thermal imager of an unmanned aerial vehicle, embodiments of the invention automatically mark the drainage lines, wires, and their maximum, minimum, and average temperatures from a single infrared image. For a plurality of pictures in the same folder, images are placed in different result folders for different temperature results of the same part; for a single image, different point markers, line markers, and rectangle markers are selected to make markers in the image. The user clicks the point mark with a mouse in the image and then sees the temperature value of the point, and after clicking for multiple times, the user sees the information of the maximum temperature, the minimum temperature, the average temperature and the like of the point in the temperature information display area. For the line mark, after a user clicks a starting point and an ending point on the image, the software automatically draws a straight line, simultaneously displays a temperature trend graph of the point on the line in a temperature trend display area, displays information such as the highest temperature, the lowest temperature and the average temperature of the line segment in a temperature information display area, and simultaneously marks specific positions of the highest temperature and the lowest temperature in the image. Similarly, for the rectangular mark, after a user clicks a starting point and an ending point on the image, the software automatically draws a rectangle, the temperature information display area displays the information of the maximum temperature, the minimum temperature, the average temperature and the like of the line segment, and the specific positions of the maximum temperature and the minimum temperature are marked in the image.
Referring to fig. 1 and 2, the high-voltage wire infrared image processing algorithm mainly includes two major processes of image preprocessing (feature extraction) and linear classification identification, wherein the preprocessing includes major processes of image channel separation, ROI extraction, edge detection, connected domain processing and the like, and each linear classification needs to perform targeted feature processing according to wire features to finally obtain a semantic segmentation region of each linear target. The method comprises the following specific steps:
s1: acquiring an infrared image through a thermal imager carried by the unmanned aerial vehicle;
s2: separating RGB channels;
s3: adaptive gray threshold processing for denoising the image;
noise often appears as isolated pixel points or pixel blocks with strong visual effect on an image, the image noise can seriously interfere the subsequent algorithm processing process, and the source of the noise generally has two modes, namely, the noise is influenced by the properties of sensor materials, working environment, electronic components, circuit structures and the like in the image acquisition process, and various noises such as thermal noise caused by resistance, channel thermal noise of a field effect tube, photon noise, dark current noise and photoresponse non-uniformity noise can be introduced; and secondly, in the preprocessing process of the image, isolated pixel points or pixel spots are formed by amplifying non-uniform intensity caused by feature enhancement. The grayed image is called Gaussian noise if the probability density function of certain noise presents a normal distribution rule, the time-varying mean value and two instantaneous covariance functions can be used for determining the Gaussian noise, and if the noise is stable, the mean value is independent of time and is equivalent to the power spectral density in meaning. Gaussian noise can be generated by a large number of independent pulses, so that each of these pulses has a value that is negligible compared to the sum of all the pulse values during any finite time interval. The stationary noise represents that the mean value is not correlated with time, while the covariance function becomes a correlation function only related to the difference between two considered instants, the function is the same as the power spectral density, and the probability density function is as follows:
besides, the noise in the linear object is mainly the interference term in the image background, such as cloud, sun, etc. After the gray values in the image are counted, the tower, the cloud and the sun in the image are layered, the gray value of the sun is the highest, and the cloud is lower in the tower. An adaptive gray-level threshold algorithm is used to perform binary processing of the image to filter the distracters in these backgrounds.
S4: extracting an ROI (region of interest) of the image; the ROI extraction mainly comprises the step of delimiting and segmenting a main object contour in a linear object, and provides a basis for the next work of image preprocessing. Several images that have been studied at this stage include threshold-based segmentation methods, edge-based segmentation methods, and segmentation methods based on cluster analysis or fuzzy set theory, among others. Clustering refers to the manner in which a group of study objects is classified into groups of units having the same characteristics. Clustering is commonly used to partition an unsigned data collection. The K-means algorithm is a more classical and common clustering method.
Linear class target recognition can apply the improved algorithm of k-mean to infrared image processing. The target infrared image is simply preprocessed, and the processed image is used as a gray image imported in the first step of the algorithm; based on a K-means algorithm, the number of clustering numbers K of the image to be iterated is set by using the gray level distribution frequency characteristics of the histogram and the number of peak-valley and flat areas. Secondly, setting the central point of the initial grouping by using a quantile point strategy, then gradually realizing the mean value iteration in each unit, finally finishing the segmentation, next calculating the optimal classification quantity by using an optimization criterion, verifying the k value preset in the first step, and exporting the segmented infrared image after satisfying the ideal segmentation effect. The standard is used for overcoming the defect of randomly selecting the clustering value and the initial clustering center, verifying a more proper clustering value and a more perfect segmentation result, and has higher practicability and reliability. The flow is shown in fig. 3.
Firstly, all gray values which are not subjected to grouping processing in a gray image after simple processing are arranged into a vector set X according to a monotone rising principle, then the set is averagely divided into k parts according to a calculated k value, the position between every two equal parts is the position of a quantile point, and the corresponding numerical value is Pi. The calculation formula is as follows:
assuming that the k value of the selected image is 6, 5 sextant numbers are obtained by sorting and bisecting, the numerical value of the first sextant number is equal to the numerical value of one sixth position of all the gray values in the set after being sorted according to the monotone ascending sequence, the numerical value corresponds to P1, and so on, the calculation modes of P2, P3, P4 and P5 are the same. In order to achieve a more ideal segmentation effect, the inner difference of each group after classification is the smallest, namely the value of each element is as close to the average value of the group, in a pixel set, and the difference between the groups is the largest, namely the difference between the average values of different groups is the better. The internal difference value of the same group is represented as the standard deviation between the pixel values of all elements in one group and the corresponding pixel value of the center point, i.e., the average value within the group. The expression of the internal difference value of the whole pixel set is:
in the formula: n is the number of all pixels in the set, x represents the pixel value of each element in the ith group, and Z is the average of all pixel values in the ith group. The results are shown in FIG. 4.
S5: extracting edges; and detecting the straight line of the image by using a Canny operator, and converting the detection of the set of the points with more significant local changes in the gray level image into edge detection. The local peak of the amplitude of the first derivative of the image grey profile corresponds to a step edge, i.e. the zero crossing of the corresponding second derivative. In numerical order, gradient is a measure used to interpret the change in function, and an image can be seen as a series of sample points where the image intensity function changes continuously. Thus, if there is a significant change in its gray value in one image, it is detected by a discrete approximation function with a gradient. The gradient appears as a two-dimensional equivalent of the first derivative, i.e.:
specifically, the substitute gradient amplitude a (x, y) = | P is approximated using an absolute value x |+|P y | the gradient direction is expressed as
a. Image smoothing
The edge detection algorithm of the image is mainly based on the first-order and second-order derivative calculation of the image gray scale, but is easily influenced by noise in the derivative calculation process, so a filter must be used firstly to inhibit the adverse influence of the noise factor on the related edge detection performance.
The Canny edge detector is an optimized approximation operator for the product of signal-to-noise ratio and localization, which derives the first derivative form of the gaussian function by using a norm derivation method. Because the convolution operation has the property of being exchangeable and combinable, the detection method firstly passes through a two-dimensional Gaussian function
The image is smoothed, so that the aim of suppressing the image noise is fulfilled.
b. Calculation of gradient magnitude and direction
After the image smoothing operation is performed, in order to obtain a corresponding gradient magnitude image and a corresponding gradient direction image, a conventional detection algorithm calculates the gradient magnitude and the direction of a smoothed data array I (x, y) by using a method of selecting each pixel point in a 2 × 2 neighborhood and calculating a finite difference mean value of the pixel points, and a partial derivative of the two directions is
G x ＝(I(i,j+1)-I(i,j)+I(i+1,j+1)-I(i+1,j))/2
G y ＝(I(i,j)-I(i+1,j)+I(i,j+1)-I(i+1,j+1))/2；
The gradient magnitude and gradient direction are calculated respectively
c. Non-maxima suppression, detection and connection edges
The Canny algorithm firstly carries out direction angle normalization operation on the gradient direction of the image, finally normalizes the direction angle normalization operation to four angles, and then inhibits the amplitude of all non-roof ridge (characteristic contour line in the gradient direction) peak values existing in the gradient direction to achieve the effect of thinning the gradient amplitude roof ridge in the amplitude array A [ i, j ]. Selecting a template with the size of a 3 multiplied by 3 pixel window and containing 8 direction neighborhoods to act on all points of the amplitude array A [ i, j ], comparing a central pixel with two pixels along a gradient line at each point, and if the amplitude A [ i, j ] at the central pixel point is not larger than the amplitudes of two adjacent points along the gradient line in the comparison process, setting the A [ i, j ] to be zero at the moment. After the processing, the wide roof ridge can be thinned to be only single-pixel-point wide, so that the width value of the roof ridge is thinned in the non-maximum value inhibition process, and the height value of the roof ridge is reserved.
The final result of the non-maximum suppression is an edge point array of the image, but various false edge information caused by noise or texture still inevitably exists in the edge array, so that a further edge thresholding operation is required to further eliminate the false edge. The results are shown in FIG. 5.
S6: selecting a Seed-Filling Seed Filling method to obtain a maximum connected domain, wherein the connected domain is the position of the insulator;
after the linear image is subjected to ROI extraction and edge detection processing, the main target object in the image can be basically subjected to region division, the position information of the insulator can be firstly determined according to the maximum connected domain by the linear image, after the position is determined, the boundary point position of the wire at the edge of the image can be estimated according to the inclination of the insulator, and finally the final wire position is determined by counting high-gray-level pixel points between the wire end point and the boundary end point at the insulator.
Since the luminance values of a binary image are only two gray values of 0 and 255, the common connected regions have 4 adjacent regions and 8 adjacent regions (as shown in fig. 6) while the basic analysis methods for the connected regions are mainly two types: two-Pass scanning method, seed-Filling Seed Filling method. We chose the Seed-Filling Seed Filling method to find the largest connected domain. The flow chart is shown in fig. 7.
The insulator and the wire have a strongly correlated topological relation in space, and the processing of the maximum connected domain of the insulator can improve the position reference for the judgment of the wire/drainage wire, so that the robustness of the linear target detection is improved. The result graph is shown in FIG. 8.
S7: identifying a lead; through the image preprocessing process, the visual characteristics of the linear targets in the infrared image are defined, the object topology of the image pixel space is calibrated, and then targeted processing can be performed according to the actual characteristics of the linear targets, so that semantic segmentation and fine extraction of target areas are completed. After studying and judging the linear target samples, the specific processing flow of the wire identification algorithm is as follows:
a. performing graying operation on the obtained infrared image to obtain a corresponding grayscale image, performing denoising and smoothing pretreatment on the grayscale image, and judging and removing ROI (region of interest) of an extracted lead and a drainage wire of an iron tower and the like through a connected domain in the grayscale image;
b. performing morphological operation on the images according to the extracted ROI area images of the conducting wires and the drainage wires to enable the insulator connected regions to be connected into a whole, and then searching the largest connected region area in the image to be the insulator part with the largest visual area in the image;
c. in the gray scale obtained by the preprocessing, after the number of wires is determined by the determined positions of the insulators, the boundary of the insulator on the wire side is taken as a starting point, and the intersection point of the wire and the image boundary is taken as an end point. And counting the number of pixel points with higher gray values on the lines. The most pixel conductive lines are the detected conductive lines.
d. Because the gray-scale image obtained by preprocessing is subjected to morphological expansion, the obtained lead information may have slight deviation with the lead information in the original image, so that the lead position is extracted, the gray-scale values of pixel points on the lead are sequentially calculated in the original gray-scale image, secondary positioning is carried out, error pixel points are eliminated, and more accurate single-pixel lead position information is obtained. The flow of the secondary positioning is shown in fig. 9, and the final algorithm detection result is shown in fig. 10.
Referring to fig. 11, the embodiment of the present invention includes a picture folder selection module, a picture saving module, a folder directory tree display module, a picture processing display module, a tag type selection module, a temperature trend display module, a temperature information display module, a picture batch processing parameter setting module, and a fast detection module.
The picture folder selection module is used for providing pictures for the following operation, and the picture folder can be selected in two modes. The first is to click a file button in the menu, then click to open a folder, select a picture folder path in the popped-up file directory, click to determine, and then see a path expansion tree in the file tree directory on the left side of the interface, see fig. 12. The second is to click directly on the left file tree directory, expanding one layer at a time until the picture is opened, see fig. 13.
And the picture saving module clicks a file button in the menu, then clicks another saving button, selects a path in the popped file directory, and can save the current picture to the position under the path after clicking the certain path.
The folder directory tree-shaped display module is used for selecting other pictures in the picture folder and selecting and displaying a picture folder path.
The image processing and displaying module is used for clicking a button for processing the image after the image is selected, and the background algorithm can automatically mark the drainage wire and the wire part in the image. In addition, clicking on the clear picture, and selecting a different marker, clicking on the picture can correspondingly display the highest, lowest, and average temperature information, see fig. 14.
The mark type module is used for matching with the picture processing and displaying module and displaying the highest, lowest and average temperature information on the picture correspondingly by clicking the mark.
The temperature trend display module is used for clicking the starting point and the ending point on the picture after the line mark is selected, and a straight line can be displayed on the picture. Meanwhile, the middle and lower areas of the software interface display the temperature distribution trend of all points on the line.
And a temperature information display module which displays the highest temperature, the lowest temperature, and the average temperature among all points when the point mark is selected and a plurality of points are clicked in the image, see fig. 15. When a line marker is selected and a plurality of lines are drawn in the image, the module displays the highest temperature, lowest temperature, average temperature for each line, and at the same time, when a different line is selected, the temperature trend for that line will be displayed in the temperature trend module, see FIG. 16. When the rectangular mark is selected, the module displays the highest temperature, the lowest temperature and the average temperature of each rectangular area. And the highest and lowest temperature locations are marked in the corresponding rectangular areas, see fig. 17.
The picture information display module is used for displaying the related information of the current picture.
Referring to fig. 18 and 20, the image batch processing parameter setting module is configured to automatically create ok, ng, and temp folders after clicking "\8230;" the button selects a path, and respectively store the images whose highest temperatures meet the sum of the warning temperature value + the threshold, the non-conforming images, and the abnormal images which do not meet the algorithm processing. Clicking on the "start batch" button then appears to the process progress bar dialog box and the cancel button can be clicked to end the batch process ahead of time, see FIG. 19.
The above embodiments are only used for illustrating the design idea and features of the present invention, and the purpose of the present invention is to enable those skilled in the art to understand the content of the present invention and implement it accordingly, and the protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes and modifications made in accordance with the principles and concepts disclosed herein are intended to be included within the scope of the present invention.
Claims (10)
1. A high-voltage wire infrared image processing method based on an infrared camera is characterized in that: the method comprises the following steps:
s1: collecting an infrared image by an unmanned aerial vehicle thermal imager;
s2: preprocessing the infrared image to determine the visual characteristics of a linear target in the infrared image, and calibrating the object topology of an image pixel space; the method comprises the following specific steps:
s21: carrying out graying operation on the infrared image to obtain a corresponding grayscale image; then preprocessing the gray level image including denoising and smoothing; then judging and removing non-target objects through a connected domain of the gray level image, and extracting ROI areas of the leads and the drainage wires;
s22: performing morphological operation on the image according to the ROI of the lead and the drainage wire to connect the insulator connected regions into a whole; then searching the largest connected domain area in the image, namely the insulator part with the largest visual area in the image;
s23: in the gray scale image, counting the number of pixel points with higher gray scale values on the lines by taking the boundary of the insulator on the side of the conductor as a starting point and the intersection point of the conductor and the image boundary as an end point according to the positions of the insulators and the number of the conductors, wherein the line with the most pixel points is the detected conductor;
s24: extracting the position of the wire, sequentially calculating the gray values of pixel points on the wire in an original gray image to perform secondary positioning, eliminating error pixel points and obtaining accurate single-pixel wire position information;
s3: and performing targeted processing according to the actual characteristics of the linear objects in the image to finish semantic segmentation and target area refinement extraction of the linear objects.
2. The infrared image processing method of the high-voltage wire based on the infrared camera as claimed in claim 1, wherein: in the step S21, the specific steps are as follows:
s211: carrying out RGB channel separation on the infrared image to obtain a gray image;
s212: performing binary processing on the gray level image by adopting a self-adaptive gray level threshold algorithm for denoising the image;
s213: and (3) extracting an ROI (region of interest) region of the denoised gray image by adopting a K-means algorithm, and being used for delimiting and segmenting the main body target profile of the linear target.
3. The infrared image processing method of the high-voltage wire based on the infrared camera as claimed in claim 2, wherein: in step S213, the specific steps are:
s2131: setting the number k of clusters of the gray image needing iteration by using the gray distribution frequency characteristics of the histogram and the number of peak valleys and slow areas;
s2132: setting the central point of the initial grouping by using a quantile point strategy, gradually realizing the mean value iteration in each unit, and finishing the segmentation;
arranging all the gray values which are not subjected to grouping processing in the processed gray image into a vector set X according to a monotone rising principle;
dividing the set X into k parts according to the set k value, and then the position between each part is the position P of the quantile i ：
S2133: calculating the optimal classification quantity by using an optimization criterion, verifying a preset k value until an ideal segmentation effect is met, and then deriving a segmented infrared image;
in a pixel set, the internal difference of each group after classification is minimum, namely the numerical value of each element is as close as possible to the average value of the group, and the difference between the groups is maximum, namely the average difference between different groups is larger, so that the segmentation effect is better;
the internal difference value of the same group is expressed as a standard deviation between the pixel values of all elements in one group and the pixel value corresponding to the central point, i.e. the average value within the group;
let n be the number of all pixels in the set, C i For the ith group, x is the pixel value of each pixel in the ith group, and Z is the average of all the pixel values in the ith group, then the internal difference value S of the whole pixel set in Comprises the following steps:
4. the infrared image processing method of the high-voltage wire based on the infrared camera as claimed in claim 2, wherein: in the step S22, the specific steps are as follows:
s221: detecting straight lines in the image by adopting a Canny operator, and converting the detection of a set of points with more significant local changes in the gray level image into edge detection;
s222: and (3) carrying out connected domain treatment by adopting a Seed-Filling Seed Filling method to obtain a maximum connected domain, wherein the connected domain is the position of the insulator.
5. The infrared image processing method of the high-voltage wire based on the infrared camera according to claim 4, characterized in that: in step S221, the specific steps are as follows:
s2211: let (x, y) be the coordinates of a certain pixel in the image, x 2 +y 2 For the fuzzy radius, σ is the standard deviation of normal distribution, a two-dimensional gaussian function is adopted to perform smoothing operation on the image to obtain a data array I (x, y) for suppressing the adverse effect of image noise factors on the edge detection performance, and the two-dimensional gaussian function is as follows:
in a two-dimensional space, the contour lines of the curved surface generated by the formula are concentric circles which are normally distributed from the center; a convolution matrix formed by pixels with non-zero distribution is transformed with the original image, and the value of each pixel is the weighted average of the values of the surrounding adjacent pixels; the value of the original pixel has a maximum Gaussian distribution value and a maximum weight, and the weight of the adjacent pixel is smaller as the distance from the adjacent pixel to the original pixel is farther;
s2212: calculating the gradient amplitude and the direction of the smoothed data array I (x, y) by selecting a method of solving finite difference mean values of all pixel points in a 2 x 2 neighborhood to obtain a corresponding gradient amplitude image and a corresponding gradient direction image;
the partial derivatives in the x-and y-directions are:
G x ＝(I(i,j+1)-I(i,j)+I(i+1,j+1)-I(i+1,j))/2
G y ＝(I(i,j)-I(i+1,j)+I(i,j+1)-I(i+1,j+1))/2；
the gradient magnitude and gradient direction are respectively:
s2213: carrying out non-maximum value inhibition, detection and edge connection by adopting a Canny algorithm to obtain an edge point array of the image, and carrying out further edge thresholding operation to remove false edges;
implementing a direction angle normalization operation in a gradient direction of the image to normalize to four angles;
the amplitude of all non-roof ridge peak values existing in the gradient direction is restrained, and the effect of thinning the gradient amplitude roof ridge in the amplitude array A [ i, j ] is achieved;
selecting a template with the size of a 3 multiplied by 3 pixel window and containing 8 direction neighborhoods to act on all points of the amplitude array A [ i, j ], comparing a central pixel with two pixels along a gradient line at each point, and setting the A [ i, j ] to zero if the amplitude A [ i, j ] at the central pixel point is not larger than the amplitudes of two adjacent points along the gradient line in the comparison process.
6. The infrared image processing method of the high-voltage wire based on the infrared camera according to claim 1, characterized in that: in the step S23, the specific steps are as follows:
s231: determining the position information of the insulator according to the maximum connected domain;
s232: estimating the position of a boundary point of a wire at the edge of an image according to the inclination of the insulator;
s233: and determining the final position of the wire by counting high-gray pixel points between the end point of the wire at the insulator and the boundary end point.
7. The infrared image processing method of the high-voltage wire based on the infrared camera according to claim 1, characterized in that: in the step S3, the specific steps are as follows:
s31: storing a plurality of images in the same folder into different result folders according to different temperature results of the same part;
s32: selecting different point marks, line marks and rectangular marks for a single image to mark in the image; the drainage lines, wires and corresponding maximum, minimum and average temperatures in a single infrared image are marked.
8. The infrared image processing method of the high-voltage wire based on the infrared camera according to claim 7, characterized in that: in the step S32, the specific steps are as follows:
s321: the method comprises the steps that point marks are marked, a user clicks the point marks in an image and then displays temperature values of the points, and after the point marks are clicked for multiple times, information including the highest temperature, the lowest temperature and the average temperature of the point marks is displayed in a temperature information display area;
s322: the method comprises the steps that a line alignment mark is marked, a user clicks a starting point and an ending point on an image to generate and display a line segment, a temperature trend graph of the points on the line segment is displayed in a temperature trend display area, information including the highest temperature, the lowest temperature and the average temperature of the line segment is displayed in a temperature information display area, and specific positions of the highest temperature and the lowest temperature are marked in the image;
s323: for the rectangular mark, a user clicks a starting point and an ending point on the image to generate and display a rectangle, information including the maximum temperature, the minimum temperature and the average temperature of the rectangle is displayed in the temperature information display area, and meanwhile specific positions of the maximum temperature and the minimum temperature are marked in the image.
9. A system for the infrared image processing method of the high-voltage wire based on the infrared camera of any one of claims 1 to 8, characterized in that: the system comprises a picture folder selection module, a picture storage module, a folder catalog tree-shaped display module, a picture processing display module, a mark type selection module, a temperature trend display module, a temperature information display module, a picture information display module and a picture batch processing parameter setting module;
the picture folder selection module is used for providing pictures for subsequent operations;
the picture saving module is used for saving the current picture according to the selected path;
the folder directory tree-shaped display module is used for selecting pictures in the picture folder according to the path;
the picture processing and displaying module is used for marking drainage wires and conducting wires and corresponding highest temperature, lowest temperature and average temperature information in the selected picture;
the mark type module is used for matching with the picture processing and displaying module and displaying corresponding maximum temperature, minimum temperature and average temperature information on the picture by clicking the mark;
the temperature trend display module is used for displaying a line segment on the picture according to the line marking instruction, the positions of the starting point and the ending point and displaying the temperature distribution trend of all points on the line segment;
the temperature information display module is used for displaying the highest temperature, the lowest temperature and the average temperature information of all the points on the picture according to the point marking instruction and the positions of the points; the temperature display device is also used for displaying the information of the highest temperature, the lowest temperature and the average temperature of each line according to the line marking instruction and the positions of a plurality of lines and displaying the temperature trends of different lines; the temperature display module is also used for displaying the highest temperature, the lowest temperature and the average temperature information of each rectangular area according to the rectangular marking instruction and the position of the rectangular area, and marking the position of the highest temperature and the lowest temperature in the corresponding rectangular area;
the picture information display module is used for displaying the related information of the current picture;
the picture batch processing parameter setting module is used for respectively storing pictures with the highest temperature meeting the sum of the early warning temperature value and the threshold value.
10. A computer storage medium, characterized in that: stored with a computer program executable by a computer processor, the computer program performing a high voltage wire infrared image processing method based on an infrared camera according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211503387.3A CN115880501A (en) | 2022-11-28 | 2022-11-28 | High-voltage wire infrared image processing method and system based on infrared camera |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211503387.3A CN115880501A (en) | 2022-11-28 | 2022-11-28 | High-voltage wire infrared image processing method and system based on infrared camera |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115880501A true CN115880501A (en) | 2023-03-31 |
Family
ID=85764354
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211503387.3A Pending CN115880501A (en) | 2022-11-28 | 2022-11-28 | High-voltage wire infrared image processing method and system based on infrared camera |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115880501A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116503633A (en) * | 2023-07-03 | 2023-07-28 | 山东艾迈科思电气有限公司 | Intelligent detection control method for switch cabinet state based on image recognition |
| CN116543238A (en) * | 2023-07-06 | 2023-08-04 | 深圳市天迈通信技术有限公司 | Image detection method for cable insulating layer |
-
2022
- 2022-11-28 CN CN202211503387.3A patent/CN115880501A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116503633A (en) * | 2023-07-03 | 2023-07-28 | 山东艾迈科思电气有限公司 | Intelligent detection control method for switch cabinet state based on image recognition |
| CN116503633B (en) * | 2023-07-03 | 2023-09-05 | 山东艾迈科思电气有限公司 | Intelligent detection control method for switch cabinet state based on image recognition |
| CN116543238A (en) * | 2023-07-06 | 2023-08-04 | 深圳市天迈通信技术有限公司 | Image detection method for cable insulating layer |
| CN116543238B (en) * | 2023-07-06 | 2023-09-01 | 深圳市天迈通信技术有限公司 | Image detection method for cable insulating layer |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110148162B (en) | Heterogeneous image matching method based on composite operator | |
| CN107610114B (en) | optical satellite remote sensing image cloud and snow fog detection method based on support vector machine | |
| EP3161787B1 (en) | Detecting edges of a nucleus using image analysis | |
| CN115880501A (en) | High-voltage wire infrared image processing method and system based on infrared camera | |
| EP3176751B1 (en) | Information processing device, information processing method, computer-readable recording medium, and inspection system | |
| EP0363828A2 (en) | Method and apparatus for adaptive learning type general purpose image measurement and recognition | |
| CN111079596A (en) | System and method for identifying typical marine artificial target of high-resolution remote sensing image | |
| CN111145223A (en) | Multi-camera personnel behavior track identification analysis method | |
| Ye et al. | Overhead ground wire detection by fusion global and local features and supervised learning method for a cable inspection robot | |
| Thilagavathi et al. | Application of image processing in diagnosing guava leaf diseases | |
| CN113447771A (en) | Partial discharge pattern recognition method based on SIFT-LDA characteristics | |
| CN111091071B (en) | Underground target detection method and system based on ground penetrating radar hyperbolic wave fitting | |
| Ibraheem et al. | A non-invasive automatic skin cancer detection system for characterizing malignant melanoma from seborrheic keratosis | |
| CN109829902B (en) | Lung CT image nodule screening method based on generalized S transformation and Teager attribute | |
| CN114581658A (en) | Target detection method and device based on computer vision | |
| CN114373079A (en) | Rapid and accurate ground penetrating radar target detection method | |
| Chen et al. | Image segmentation based on mathematical morphological operator | |
| CN113344047A (en) | Platen state identification method based on improved K-means algorithm | |
| EP1194880A1 (en) | Computer aided detection of masses and clustered microcalcification strategies | |
| CN117292131A (en) | Mass spectrum peak detection algorithm | |
| CN109785318B (en) | Remote sensing image change detection method based on facial line primitive association constraint | |
| CN108805186B (en) | SAR image circular oil depot detection method based on multi-dimensional significant feature clustering | |
| Walker et al. | Locating salient facial features using image invariants | |
| CN111047616A (en) | Remote sensing image landslide target constraint active contour feature extraction method | |
| Denuelle et al. | Kelp detection in highly dynamic environments using texture recognition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |